Gene Disruption by Homologous Recombination in the Xylella fastidiosa Citrus Variegated Chlorosis Strain

Abstract

Mutagenesis by homologous recombination was evaluated in Xylella fastidiosa by using the bga gene, coding for β-galactosidase, as a model. Integration of replicative plasmids by homologous recombination between the cloned truncated copy of bga and the endogenous gene was produced by one or two crossover events leading to β-galactosidase mutants. A promoterless chloramphenicol acetyltransferase gene was used to monitor the expression of the target gene and to select a cvaB mutant.

Xylella fastidiosa is a fastidious, gram-negative, xylem-limited bacterium (20) that causes a range of economically important plant diseases, including citrus variegated chlorosis (CVC) (2, 17); Pierce's disease (PD) of grapevine; alfalfa dwarf; leaf scorch of almond, coffee, elm, sycamore, oak, plum, mulberry, maple, and oleander; and periwinkle wilt (15, 16). Despite the importance of the X. fastidiosa CVC strain in phytopathology, our understanding of the physiology and genetics of this bacterium is still poor. Genetic tools to study the biology of X. fastidiosa are limited due to the difficulty in culturing and transforming this fastidious organism. Production of mutants is an important and necessary way to identify and study genes and then the mechanisms involved in different processes, such as pathogenicity. Several methods could be used to produce mutants: insertion-duplication mutagenesis (IDM), allelic exchange (AE), and transposon mutagenesis. Recently, random mutagenesis by transposition was used to produce mutants in X. fastidiosa PD strains (7). However, this method could not be used to directly inactivate specific genes, which is achieved by homologous recombination (IDM and AE). IDM has been used to disrupt genes in a variety of other organisms, such as Mycobacterium smegmatis (1), Neisseria gonorrhoeae (8), Streptococcus pneumoniae (12), and Lactobacillus sake (13). This mutagenesis involves circular integration, by a single crossover event, between the targeted chromosomal gene and a truncated copy of this gene cloned in a transient suicide or replicative plasmid, resulting in integration of the entire plasmid and duplication of the target sequence. AE results in the replacement of the endogenous gene by its copy disrupted by a selectable marker. This approach involves homologous recombination with two crossovers. We previously reported the transformation of X. fastidiosa with artificial plasmids carrying the X. fastidiosa chromosomal origin of replication (oriC) and a kanamycin resistance gene under the control of the X. fastidiosa 16S rRNA promoter (14). These X. fastidiosa oriC plasmids were found to be integrated in the chromosome at the rRNA promoter site by homologous recombination involving one crossover (14), suggesting that gene disruption by homologous recombination is possible in X. fastidiosa. Here we report the disruption of genes by homologous recombination involving one crossover (IDM) or two crossovers (AE) as tools to produce mutants in X. fastidiosa by site-directed gene disruption.

Construction of plasmid for IDM in X. fastidiosa.

The plasmid used for IDM (pBCK492) (Fig. 1A) was derived from pBS. Since transformation with suicide plasmids was not achieved for X. fastidiosa CVC (14) or PD (7) strains in previous studies, we used replicative plasmids that offer different advantages over the use of nonreplicating plasmids. (i) They allow the dissociation of transformation from recombination efficiencies. (ii) The continued presence of replicative plasmid in the bacterial cell allows detection of very rare recombination events. (iii) The use of replicative plasmids should increase the recombination rate, since homologous recombination has been suggested to occur during DNA replication (19). Replication of plasmids into X. fastidiosa was obtained by cloning the 366-bp BamHI fragment from pBKori (14), corresponding to the oriC of X. fastidiosa strain 9a5c, into pBS (Fig. 1A). The aacA-aphD gene coding for kanamycin resistance, carried by the PstI fragment from pUC4K, was cloned and used as a selection marker for transformation. For homologous recombination with a suicide plasmid, transformant and mutant clones were selected by their resistance to a marker introduced into the chromosome by the plasmid. In contrast, the selection for transformation and integration by a replicative plasmid requires a recombination marker different from the transformation marker. For that purpose, a promoterless chloramphenicol acetyltransferase (CAT) gene (Fig. 1A), carried by a HindIII fragment from plasmid pKPFCAT (a gift from Joel Renaudin, INRA, Bordeaux Research Center, Bordeaux, France), was transcriptionally fused to the truncated gene and used as a marker to select the recombinant clones.

FIG. 1.

(A) Schematic representation and partial restriction map of pBCK492 used for IDM. (B) Sizes and positions of the truncated fragments of the bga gene used for IDM. (C) Plasmids pB42galK5, pB42galK6, and pBgalK used for AE.

Production of X. fastidiosa β-galactosidase-deficient mutants by IDM.

In order to test the capacity of gene disruption by homologous recombination to produce X. fastidiosa mutants, we chose the bga gene coding for β-galactosidase, because of its detectable phenotype. The genome sequence of X. fastidiosa CVC strain 9a5c revealed one open reading frame (XF0840) (18) predicted to code for a β-galactosidase enzyme. The putative protein coded by the X. fastidiosa bga gene shows similarity to β-galactosidases from Xanthomonas campestris pv. manihotis (80%, accession no. P48982), Streptococcus pneumoniae (56%, accession no. NP_357653), Arabidopsis thaliana (54%, accession no. AAL25611), and Homo sapiens (53%, accession no. AAA51822). However no similarity was observed with β-galactosidase from Escherichia coli. Analyses of the deduced amino acid sequence by using ProDom (3) revealed that this putative enzyme belongs to family 35 of glycosyl hydrolases (9, 10). The consensus sequence pattern, used as a signature sequence and presumed to be the active site, is conserved in the putative enzyme from X. fastidiosa CVC strain 9a5c. These observations suggest that the putative β-galactosidase from strain 9a5c is functional. To detect the predicted β-galactosidase activity, X. fastidiosa CVC strains 9a5c and J1a12 (14) were plated onto PWG agar medium (PW medium [4] supplemented with 0.5% glucose) with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) as a substrate. We verified that the blue coloration of colonies from β-galactosidase activity was only observed in the absence of phytone peptone, one of the components of PWG medium. This result suggests a regulation of X. fastidiosa β-galactosidase activity by phytone peptone. Because the minimal length of homology needed to detect homologous recombination in X. fastidiosa is unknown, we amplified by PCR several truncated fragments of bga with deletions in the 5′ and 3′ ends having sizes ranging from 190 to 1,616 bp (Fig. 1B and Table 1). The truncated fragments of the bga gene were cloned in plasmid pBCK492, generating the recombinant plasmids pBCKgal1R1, pBCKgal1R2, pBCKgal1R3, pBCKgal1R4, pBCKgal1R5, and pBCKgal2R5. Plasmid pBCKgal1R6, different from the others, carries a truncated copy of the bga gene with a unique deletion in the 5′ end (Fig. 1B). The integration of this plasmid by homologous recombination would lead to the creation of two copies of bga: a complete copy under control of its endogenous promoter and a second copy with deletion at the 5′ end. The recombinant plasmids were used to transform strain J1a12 by electroporation as previously described (14). The transformation frequency ranged from 1 × 10⁻⁷ to 7 × 10⁻⁷ per μg of plasmid DNA when pBCK492 was used as a control. After 3 weeks of incubation, five colonies from each transformation experiment were cultivated in liquid PWG medium supplemented with 5 μg of kanamycin per ml. The presence of the recombinant plasmids in the transformants after six passages (one passage corresponds to the growth of Xylella for 8 days from a 1/10 dilution of a stationary-growth-phase culture into a new medium) in PWG liquid medium containing 5 μg of kanamycin per ml was verified by PCR with specific primers for the CAT gene. Integration of plasmids into the recipient chromosome was determined by Southern blot hybridization (data not shown). After six passages in liquid medium, only free plasmid could be detected in 26 cultures out of 35. Four cultures out of 35 showed free plasmid as well as integrated plasmid into the chromosome, and in the 5 remaining cultures, we detected only the integrated form. These observations indicated that integration of the recombinant plasmid into the bacterial chromosome could occur earlier than six passages (i.e., earlier than 20 generations after the selection of the transformants on plates, assuming that one passage corresponds to 3.25 generations).

TABLE 1.

Inactivation of the bga gene by IDM

Recombinant plasmid used for transformation	Size of truncated bga fragment (bp)	Clone	Integration sitea		Detection of β-galactosidase activityb
			6th passage	12th passage
pBCKgal1R1	190	Gal1R1.1	oriC	oriC	+
		Gal1R1.2	—	—	+
		Gal1R1.3	—	ND	ND
		Gal1R1.4	oriC (p)	oriC	+
		Gal1R1.5	oriC (p)	oriC	+
pBCKgal2R5	373	Gal2R5.1	oriC	oriC	+
		Gal2R5.2	bga (p)	bga (p)	+/−
		Gal2R5.3	bga	bga	ND
		Gal2R5.4	bga	bga	−
		Gal2R5.5	bga	bga	−
pBCKgal1R2	521	Gal1R2.1	—	bga (p)	ND
		Gal1R2.2	—	bga	−
		Gal1R2.3	—	—	+
		Gal1R2.4	—	bga (p)	ND
		Gal1R2.5	—	—	+
pBCKgal1R3	895	Gal1R3.1	—	bga (p)	ND
		Gal1R3.2	—	bga (p)	ND
		Gal1R3.3	—	bga (p)	ND
		Gal1R3.4	—	bga (p)	ND
		Gal1R3.5	—	bga (p)	ND
pBCKgal1R4	1,025	Gal1R4.1	—	bga (p)	ND
		Gal1R4.2	—	bga (p)	ND
		Gal1R4.3	—	bga (p)	ND
		Gal1R4.4	—	bga (p)	ND
		Gal1R4.5	—	bga (p)	ND
pBCKgal1R5	1,366	Gal1R5.1	bga (p)	bga	−
		Gal1R5.2	—	ND	ND
		Gal1R5.3	—	bga (p)	ND
		Gal1R5.4	—	bga	−
		Gal1R5.5	—	bga	−
pBCKgal1R6	1,616	Gal1R6.1	—	bga (p)	+
		Gal1R6.1	—	bga (p)	+
		Gal1R6.2	—	bga (p)	+
		Gal1R6.3	—	bga (p)	+
		Gal1R6.4	—	bga (p)	+
		Gal1R6.5	—	bga (p)	+

^aoriC, integration of the recombinant plasmid into the origin of replication; bga, integration of the recombinant plasmid into the bga gene; —, integration of the recombinant plasmid not detected; p, partial integration of the plasmid in the population (cells containing the plasmid integrated and cells containing the plasmid as free replicon).

^b+ and −, presence and absence of blue colonies, respectively; +/−, presence of blue and white colonies.

^cND, not determined.

Recombinant plasmids used for transformation in this work carry the chromosomal origin of replication of X. fastidiosa and a truncated copy of a candidate gene for disruption. Thus, there are two potential integration sites in the endogenous X. fastidiosa chromosome. The integration sites of the plasmids were determined by Southern blots with a bga or oriC probe. Figure 2 shows Southern blot hybridization between a bga gene probe and total genomic DNA from clones gal1R1.1, gal1R2.3, gal2R5.3, gal2R5.2, and gal1R5.1 (Table 1). The 4.0-kbp SphI fragment corresponding to the endogenous bga gene and the 6.0-kbp fragment corresponding to the free plasmid were not observed for clone gal2R5.3 (Fig. 2, lane 4). Instead, this clone presented two different restriction fragments of 3.0 and 7.0 kbp carrying the bga sequence. This result shows that pBCKgal2R5 is integrated by homologous recombination in the bga gene in clone gal2R5.3. In clones gal2R5.2 and gal1R5.1 (Fig. 2, lanes 5 and 6, respectively), four restriction fragments are observed: (i) two DNA fragments with a weak intensity have the same sizes as the ones from the parental strain (4.0 kbp) and the plasmids (6.0 kbp for pBCKgal2R5 and 7.0 kbp for pBCKgal1R5; Fig. 2, lanes 9 and 10, respectively), and (ii) two fragments with a strong intensity (3.0 and 7.0 kbp for gal2R5.2 and 3.0 and 8.0 kbp for gal1R5.1) differ from the fragment produced by the parental strain J1a12 (Fig. 2, lanes 3 and 11) and by the corresponding plasmids used for the production of clones gal2R5.2 and gal1R5.1 (Fig. 2, lanes 9 and 10, respectively). This result confirms the integration of pBCKgal2R5 and pBCKgal1R5 by homologous recombination into the bga gene. These results indicate that the bacterial populations in the cultures of clones gal2R5.2 and gal1R5.1 are heterogeneous, with the majority of cells integrated the plasmid by homologous recombination into the bga gene, but with the remaining minor percentage of cells carrying the plasmid as free replicon. In clone gal1R2.3 (Fig. 2, lane 2), only restriction fragments corresponding to the endogenous gene and to the free plasmid are observed, indicating that pBCKgal1R2 has not integrated. In clone gal1R1.1 (Fig. 2, lane 1), the DNA fragment corresponding to the endogenous gene and a fragment of 8.0 kbp are observed, but the DNA fragment corresponding to the free plasmid is absent (Fig. 2, lane 1). These results indicate that the plasmid is integrated in another place in the chromosome, probably at the oriC. To investigate this hypothesis, Southern blots hybridized with oriC probe were conducted (Fig. 3). For clones gal1R1.2 (lane 2), gal2R5.3 (lane 4), and gal2R5.4 (lane 5), only the fragment corresponding to the endogenous oriC in the chromosome and the corresponding copy in the plasmid integrated into bga were observed. For clone gal1R1.1 (lane 1), only two fragments of 1.1 and 7.0 kbp were observed. The fragment of 5.8 kbp produced by the free plasmid was not observed (Fig. 3, lane 6), nor was the 2.3-kbp band produced by the nontransformed parental strain J1a12 (Fig. 3, lane 8). These data indicate that the plasmid is integrated into the chromosome by homologous recombination at the oriC in clone gal1R1.1. This result could be explained by the size of the truncated copy of the bga gene cloned in pBCKgal1R1, which is smaller than the size of the oriC fragment. In clone gal2R5.1 transformed with pBCKgal2R5, in which the size of the truncated bga gene (373 bp) is almost the same as the size of the oriC fragment (366 bp), Southern blots hybridized with oriC probe show that the plasmid is integrated into the oriC region in the chromosome (Fig. 3, lane 3). In the remaining clones, the plasmid was integrated by homologous recombination at the bga gene (Table 1).

FIG. 2.

Southern blot hybridization between bga probe and SphI-restricted DNA extracted from clones gal1R1.1 (lane 1), gal1R2.3 (lane 2), gal2R5.3 (lane 4), gal2R5.2 (lane 5), and gal1R5.1 (lane 6); plasmids pBCKgal1R1 (lane 7), pBCKgal1R2 (lane 8), pBCKgal2R5 (lane 9), and pBCKgal1R5 (lane 10); and untransformed parental strain J1a12 (lanes 3 and 11).

FIG. 3.

Southern blot hybridization between the oriC probe and SphI-restricted DNA extracted from clones gal1R1.1 (lane 1), gal1R1.2 (lane 2), gal2R5.1 (lane 3), gal2R5.3 (lane 4), and gal2R5.4 (lane 5); plasmids pBCKgal1R1 (lane 6) and pBCKgal2R5 (lane 7); and untransformed parental strain J1a12 (lane 8).

After 12 passages in liquid medium supplemented with kanamycin, integration of the plasmids into the chromosome was observed for most of the transformants (30 cultures among 33) (Table 1). This suggests that the size of the truncated copy of the target gene cloned in pBCK492 determines the site of integration of the recombinant oriC plasmid: when the size of the truncated gene is smaller than the size of the oriC fragment in pBCK492 (the case in pBCKgal1R1 carrying a 190-bp fragment), the integration occurs only into the origin of replication of the chromosome, indicating a very low frequency for the integration at the endogenous bga gene or a minimum length of the target gene to allow homologous recombination in X. fastidiosa. However, when the size of the truncated gene is similar to the size of the cloned oriC (373 bp in pBCKgal2R5), the integration occurs preferentially into the target endogenous bga gene. These results could be explained by a site dependency on the recombination frequency. Differences in growth and bacterial cell or colony morphologies were not observed between the clones in which the plasmid is free or integrated into the X. fastidiosa chromosomal origin of replication or the endogenous bga gene. These observations lead us to propose that the tandem duplication of the origin of replication in X. fastidiosa seems not to modify the in vitro fitness of the bacteria. When the size of the truncated copy is bigger than the oriC fragment, the integration occurs only at the endogenous gene targeted for the disruption. These observations show that unwanted integration of IDM plasmid into oriC can be avoided by including a cloned target truncated gene that is larger than the cloned oriC.

In order to prove that the bga gene is disrupted by insertion of the recombinant plasmid (Table 1) in clones showing the recombinant organization, β-galactosidase activity was tested for these transformed clones after 12 passages in PWG liquid medium supplemented with kanamycin. Using PWG agar (without phytone peptone) supplemented with X-Gal as a substrate, blue coloration of the colonies was observed for all gal1R1 clones in which the plasmid is free or integrated at oriC (Table 1). Clones showing the integration of the plasmid at the bga gene produced only white colonies. Finally, blue and white colonies were observed for clone gal2R5.2, which had a heterogeneous population (i.e., a mixture of bacteria either carrying the plasmid as a free replicon or integrated at the bga gene) (results not shown). These results clearly show the correlation between the recombinant organization of bga and the absence of the β-galactosidase activity (i.e., the disruption of the bga gene). Only blue colonies were observed for the wild-type strain J1a12 as well as for all gal1R6 clones that contain one complete copy of the bga gene, under the control of its endogenous promoter, and a truncated copy of bga with deletion at its 5′ end.

Use of the CAT gene as a reporter for IDM in X. fastidiosa.

The promoterless CAT gene was transcriptionally fused to the truncated copy of the bga gene cloned in plasmid pBCK492 (Fig. 1A). The CAT gene could be used to select transformed clones, in which the recombinant plasmid is integrated by homologous recombination into the target gene, or to monitor the expression of the endogenous gene. The MIC of chloramphenicol for strain J1a12, transformed with pBCK492, was <5 μg/ml in PWG agar medium. In order to demonstrate the use of the CAT gene as a reporter system in X. fastidiosa, we determined the chloramphenicol resistance for clone gal1R3.3, which was randomly selected. In this clone, the CAT gene is under the control of the endogenous bga promoter because of the integration of pBCKgal1R3 at the chromosomal bga gene (Table 1). Clone gal1R3.3 was plated on PWG agar medium containing or not containing phytone peptone. As showed before, β-galactosidase was not detected in the presence of phytone peptone. Colonies of clone gal1R3.3 resistant to 5 μg of chloramphenicol per ml were observed in PWG agar medium deprived of phytone peptone, but not in the regular medium containing phytone peptone. No chloramphenicol-resistant colonies of J1a12 transformed with the control plasmid pBCK492 were obtained on PWG agar plates containing or not containing phytone peptone and supplemented with 5 μg of chloramphenicol per ml. These results, which show that the CAT gene is controlled by the promoter of the target gene, indicate that CAT can be used as a reporter for disrupted genes in X. fastidiosa.

Use of chloramphenicol to select mutants produced by site-specific gene disruption in X. fastidiosa.

To demonstrate the use of IDM to disrupt genes potentially involved in the pathogenicity of the X. fastidiosa CVC strain and the use of chloramphenicol to perform mutant selection of genes for which the phenotype is not easily detectable, we chose to disrupt the cvaB gene, which is predicted to code for one component of the colicin V ABC transporter (18). We chose cavB as a model for disruption because of its potential involvement in the virulence of X. fastidiosa (6, 11, 18) and its constitutive expression in PWG medium (Gaurivaud et al., unpublished data). Strain J1a12 was transformed with plasmid pBCKABC carrying a 1,146-bp truncated copy of cvaB. Transformants selected on PWG agar plates supplemented with kanamycin were further cultured in liquid medium containing kanamycin. The integration of pBCKABC was detected for the 12th passage. Southern blot analysis showed that the corresponding transformed clone is heterogeneous (i.e., contains cells carrying the plasmid as a free replicon and cells carrying the plasmid integrated into the cvaB gene) (data not shown). In order to test the ability to select mutants with chloramphenicol resistance, the culture was plated on PWG agar plates containing 10 μg of kanamycin per ml and supplemented or not with 5 μg of chloramphenicol per ml. Six clones (ABC1 to -6) isolated from the plates without chloramphenicol were analyzed by Southern blotting with a cvaB-specific probe (Fig. 4, lanes 1 to 6). In clones ABC1, -2, -3, -4, and −5, PstI fragments of 5.9 and 6.0 kbp corresponding to the disrupted genomic organization for the cvaB gene were observed (Fig. 4, lanes 1 to 5). However, in clone ABC4 (Fig. 4, lane 4), the 1.4-kbp restriction fragment corresponding to the free plasmid and a 13.4-kbp fragment carrying the endogenous wild-type cvaB gene were observed. In clone ABC6 (Fig. 4, lane 6), only restriction fragments corresponding to the free plasmid and to the endogenous cvaB gene are detected. Nine clones isolated from PWG agar plates supplemented with 5 μg of chloramphenicol per ml were analyzed in the same way (ABC7 to -15; Fig. 4, lanes 10 to 18). All nine clones presented the disrupted genomic organization for the cvaB gene, in which neither the endogenous gene nor the free plasmid was detected. The phenotypic characterization of X. fastidiosa cvaB-disrupted mutant is under way. These results clearly show the possibility of producing mutants by disrupting genes potentially involved in the pathogenicity of X. fastidiosa and the use of chloramphenicol for the selection of mutants produced by IDM. This system is being used to produce mutants disrupted in genes potentially involved in pathogenesis and virulence of X. fastidiosa.

FIG. 4.

Southern blot hybridization between the cvaB probe and PstI-restricted DNA extracted from clones ABC1 to -6 selected on PWG plates without chloramphenicol (lanes 1 to 6), plasmid pBCKABC (lane 7), untransformed parental strain J1a12 (lanes 8 and 9), and clones ABC7 to -15 selected on PWG plates supplemented with 5 μg of chloramphenicol per ml (lanes 10 to 18).

Analysis of the stability of oriC plasmid integration.

To assess the stability of oriC plasmid integration in X. fastidiosa, transformants were propagated in liquid medium with and without selection pressure for many generations and analyzed by PCR with specific primers that hybridize in the plasmid sequence and downstream of the integration site in the chromosome. To accomplish this, p16KdAori (14) transformants were propagated in PW liquid medium with and without kanamycin for 384 generations (16 passages), analyzed by PCR with M13 universal and Rop3 primers (14), and plated on 1% PW agar containing 20 μg of kanamycin per ml. The 2-kb fragment corresponding to the rop-Kan^r fragment was detected in all of the clones (data not shown), and no reversion to kanamycin sensitivity was observed, regardless of the presence or absence of kanamycin as the selection pressure during the serial propagations. These data demonstrate the in vitro stability of the plasmid integration over hundreds of generations. Although the integration of the oriC plasmid into the chromosome was proven to be stable in the absence of a selection pressure in the medium, reversion of such mutants to the wild-type genotype is possible in planta. To avoid this problem, we tested mutagenesis by allelic exchange in X. fastidiosa as the best way to produce stable mutants.

Disruption of the bga gene by AE.

In an attempt to investigate the capacity to disrupt genes in X. fastidiosa by AE, we have constructed plasmids pB42galK5 and pB42galK6 (Fig. 1C), carrying the origin of replication of X. fastidiosa and a truncated copy of bga gene disrupted by the insertion of the aacA-aphD gene. We also constructed, as a control, the suicide plasmid pBgalK (Fig. 1C), which carries the truncated bga gene disrupted by the aacA-aphD insertion. The wild-type strain J1a12 was transformed with these plasmids, and transformants were selected on PWG agar plates supplemented with 5 μg of kanamycin per ml. No colonies were observed when the suicide plasmid pBgalK was used. Indeed, colonies were obtained by transformation with the replicative plasmids pB45galK5 and pB42galK6. Nine colonies obtained from the transformation with each plasmid were cultivated until the 6th and 15th passages in PWG liquid medium supplemented with 5 μg of kanamycin per ml. The integration of the recombinant plasmids into the chromosome in 16 cultures out of 18 was analyzed by Southern blotting, showing that the plasmids were integrated by one crossover after six passages in PWG liquid medium supplemented with 5 μg of kanamycin per ml (data not shown). Nine passages later, the 9.0-kbp restriction fragment corresponding to the bga gene disrupted by the whole plasmid was observed for all clones (Fig. 5A). However, one SstII fragment hybridizing with the bga probe and with a size of 4.2 kbp was detected for clone galK5.5 (Fig. 5A, lane 8). The size of this fragment indicates that the endogenous bga gene is disrupted by the aacA-aphD marker. In addition, the β-galactosidase activity was not detected for this clone. Since we have observed that clone galK5.5 has a heterogeneous population (results not shown), with cells containing the bga gene disrupted by AE and cells with the bga gene disrupted by IDM, ampicillin resistance and kanamycin resistance were studied for this clone as well as for clones galK6.6 and galK6.9, which have the bga gene disrupted by a single crossover event (Fig. 5A, lanes 12 and 15, respectively). Three hundred colonies of each clone were replicated to PWG agar plates supplemented with either 5 μg of kanamycin per ml or 25 μg of ampicillin per ml. For clones galK6.6 and galK6.9, all colonies showed resistance to both antibiotics. However, for galK5.5 (Fig. 5A, lane 8), only 40% of the colonies resistant to kanamycin were resistant to ampicillin, confirming the absence of pBS in the remaining 60% of the kanamycin-resistant colonies and thus the disruption of the bga gene by AE. The genomic organization of bga (Fig. 5B) and determination of the presence of the kanamycin resistance gene (Fig. 5C) and pBS sequence (Fig. 5D) were analyzed by Southern blot hybridization for 14 isolates of clone galK5.5 showing ampicillin susceptibility and kanamycin resistance (lanes 4 to 17) and for two isolates of clone galK5.5 showing resistance to both antibiotics (lanes 18 and 19). Only one SstII restriction fragment of 4.2 kbp was observed in all isolates showing kanamycin resistance and ampicillin susceptibility (Fig. 5B, lanes 4 to 17). The size of the fragment indicates that the disruption of bga by the kanamycin resistance gene had occurred by AE. This fragment was detected in the DNA of these isolates (Fig. 5C, lanes 4 to 17) by the kanamycin probe. As expected, the pBS sequence was not detected in these isolates (Fig. 5D, lanes 4 to 17). Only one restriction fragment of 9.0 kbp hybridizing with the bga probe was detected in isolates showing resistance to both antibiotics (Fig. 5B, lanes 18 and 19). The size of this fragment agrees with the insertion of pB42galK5 by homologous recombination with one crossover in the bga gene. This conclusion is supported by the detection of the kanamycin resistance gene and the pBS sequence in these clones (Fig. 5C and D, lanes 18 and 19). These results clearly showed the occurrence of a double crossover event in X. fastidiosa and that it is possible to produce mutants through specific gene disruption by AE in this bacterium. However, the frequency of the second crossover event seems to be very low in X. fastidiosa, explaining our inability to transform X. fastidiosa with suicide plasmids. As mentioned above, Southern blot hybridization showed that the sequence of the pBS plasmid was not detected in cells that had the bga gene disrupted by AE. After the double crossover event, the remaining vector still maintains the cloned chromosomal origin of replication of X. fastidiosa, indicating that without antibiotic selection, the free oriC plasmid is not maintained in the cells. An improved version of the oriC plasmids for use to disrupt genes by AE is being constructed. A strong selection system, as described by Donnenberg and Kaper (5) using the sacB gene, might improve the frequency of double crossover in X. fastidiosa.

FIG. 5.

(A) Southern blot hybridization between the bga probe and SstII-restricted DNA from clones galK5.0 to galK5.5 (lanes 3 to 8), clones galK6.3 to galK6.10 (lanes 9 to 16), plasmid pB42galK5 (lane 1), and untransformed parental strain J1a12 (lane 2). (B to D) Southern blot hybridization between the bga (B), kanamycin (C), and pBS (D) probes to SstII-restricted DNA from 14 isolates of clone galK5.5 showing kanamycin resistance and ampicillin susceptibility (lanes 4 to 17), two isolates of clone galK5.5 showing resistance to both antibiotics (lanes 18 and 19), plasmid pB42galK5 (lane 1), and untransformed parental strain J1a12 (lanes 2 and 3).